Cloning and Characterization of PRAX-1

Using a cytoplasmic domain of the peripheral benzodiazepine receptor (PBR) as a bait in the yeast two-hybrid system, we have isolated a cDNA encoding a new protein that specifically interacts with PBR. We named it PRAX-1, for peripheral benzodiazepine receptor-associated protein 1. PRAX-1 is a 1857-amino acid protein, the sequence of which was structurally unrelated to any known proteins. The gene encoding PRAX-1 is located in the q22-q23 region of the long arm of the human chromosome 17. The PRAX-1 mRNA is 7.5 kilobase pairs, predominantly expressed in the central nervous system, pituitary gland, and thymus. At the protein level, we found the PRAX-1 as a single 220–250-kDa protein in the brain and in many different human cell lines tested using specific antibody raised against PRAX-1. Parallel analysis of the PRAX-1 mRNA and protein expression performed in mouse and rat gave similar results. Immunocytochemistry analysis carried out to define the distribution of the PRAX-1 protein in the rat brain showed that PRAX-1 was prevalent in the mesolimbic system, specially abundant in the CA1 subfield of the hippocampus. Exhibiting several domains involved in protein-protein interaction (three proline-rich domains, three leucine-zipper motifs, and an Src homology region 3-like domain), the PRAX-1 may be looked upon as a new adaptator protein. We show that both the Src homology region 3-like domain and a proline-rich domain in PRAX-1 are required for the interaction with PBR. PRAX-1 is a cytoplasmic protein that also partially colocalizes with PBR in the mitochondria, as determined by confocal microscopy and Western blotting. Altogether our observations support a model of interaction implicating PBR and this newly described protein, PRAX-1. As being the first cytoplasmic protein associated with PBR, PRAX-1 is a new tool that opens new fields for exploring PBR biological roles.

Benzodiazepines are among the most widely prescribed drugs due to their pharmacological actions in relieving anxiety, as anticonvulsant, muscle relaxant, or sedative hypnotics. They are first known to elicit their tranquilizing effect in the central nervous system via a specific receptor complex on ␥-aminobutyric acidgated chloride channels restrictedly expressed on neuronal membranes (1,2). In addition to these central type benzodiazepine receptors, a second class of binding sites was also identified in peripheral tissues and termed the peripheral benzodiazepine receptor (PBR). 1 PBR pharmacologically differs from its central counterpart in its lack of coupling to ␥-aminobutyric acid receptors and in its ligand specificity. In contrast to the central receptor, PBR exhibits nanomolar affinity to the benzodiazepine Ro5-4864 or to the isoquinoline carboxamide derivative PK11195 and low affinity for the benzodiazepine clonazepam. PBR is a 169-amino acid protein with five transmembrane domains associated with the mitochondrial outer membranes (3)(4)(5)(6)(7). PBR tissue distribution analysis revealed an ubiquitous expression of the protein with a particularly high abundance in steroid-producing tissues such as adrenal, testis, ovary, and glia. PBR is also abundant in lung, liver, kidney, and salivary glands. On the other hand, PBR expression is relatively low in skeletal muscle, gastrointestinal tract, and neurons (5). In the immune system, PBR was shown to be expressed among all human peripheral blood leukocyte subsets, mainly in monocytes and polymorphonuclear cells, and less abundantly in natural killer cells and B and T lymphocytes (8,9). To better understand PBR functions, the protein has been the object of considerable biochemical and pharmacological investigations. However, the functional assignment of PBR still has to be discovered. To date, the best documented role for PBR has been shown in steroidogenic tissues, where PBR mediates the intramitochondrial cholesterol transport, which is the rate-limiting step in steroid biosynthesis. It was found to be implicated in the acute stimulation of steroid biosynthesis, increasing pregnenolone formation (10,11). The disruption of the PBR gene in the R2C Leydig tumor cell line, a typical steroidogenic system, resulted in a dramatic decrease in the steroid biosynthesis (12). However, PBR is also expressed in cells that do not synthesize steroid; thus, the function of PBR could not be restricted to the steroidogenic field. Indeed PBR has also been implicated via specific ligands in a number of other unrelated cellular phenomena. A variety of effects of benzodiazepines on cell growth and differentiation mediated via the peripheral benzodiazepine receptor has been reported, including the following diverse actions: Ro5-4864 enhances melanogenesis in melanoma cells (13), and benzodiazepines induce the synthesis of hemoglobin in Friend erythroleukemia cells (14), inhibit proliferation of thymoma cells (15), stimulate monocytes chemotaxis (16), inhibit natural killer cell activity (17), and facilitate expression of the proto-oncogene c-fos by nerve growth factor in PC12 cells (18). Recently, another field of interest * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF039571.
¶ To whom correspondence should be addressed. Fax: 33-4-67-10-60-00; E-mail: pierre.casellas@sanofi.com. emerged considering the link between PBR and mitochondria. Mitochondria are strongly considered as key actors in apoptosis. Indeed, mitochondria were demonstrated to be implicated in the early steps of apoptosis through the alterations in the mitochondrial permeability transition, the release of the apoptosis-inducing factor sequestered in the intermembrane space of mitochondria, and in the regulation of apoptosis via the presence of Bcl-2 in the outer mitochondrial membrane (19,20). Considering the pivotal role of mitochondria in apoptosis, PBR looked upon as an actor in apoptosis must be questioned. In accordance with this, we have recently demonstrated that PBR is involved in the protection of hematopoietic cells against apoptosis following H 2 O 2 treatment (9). In this study, the expression of PBR and the resistance to H 2 O 2 toxicity on hematopoietic cell lines were found to be correlated and resistance of cells to H 2 O 2 significantly increased by PBR cDNA transfection. Thus, participating in an antioxidant pathway, PBR may play a critical role in the regulation of apoptosis events.
Previous phylogenic studies reinforced this novel idea of a PBR implicated in antioxidant pathway and modulation of apoptosis. Significant similarity exists between PBR and the CrtK protein of Rhodobacter capsulatus, a photosynthetic bacterium. The 35% identity between these proteins reveals a strong conservation of sequence between two proteins that diverged 2 billion years ago (21). This homology suggests a conserved or a highly specialized function for PBR. Indeed, the rat PBR was recently demonstrated to complement in Rhodobacter sphaeroides a mutant that lacks the tryptophanrich sensory protein (TspO) previously reported as CrtK (22). In this study, PBR was found to substitute for TspO by negatively regulating the expression of photosynthesis genes in the presence of oxygen. It is interesting that this study also suggests a pivotal role played by PBR in oxygen-dependent signal transduction.
Despite the abundance of the aforementioned PBR-mediated effects and because of their diversity, the question of the physiological role of the PBR yet remains unanswered. The key to a better understanding of the role of the PBR may be the elucidation of the molecular scheme involving PBR.
PBR has been described to belong to a complex including two different proteins of 32 and 30 kDa. These proteins have been identified as the voltage-dependent anion channel and the adenine nucleotide carrier, respectively (23). With the exception of this complex, no other protein-protein interaction implicating PBR has been established so far. Therefore, we set out to identify proteins that interact with PBR using the yeast twohybrid system (24) in the hope that this might shed a light on the roles of PBR in the cellular metabolism. Searching cDNA libraries by the two-hybrid system made it possible to identify proteins that specifically interact with our bait PBR and to clone the corresponding encoding genes. However, as a nuclearbased system, the two-hybrid system was mainly used to detect interactions between soluble proteins. Indeed if the bait protein exhibits transmembrane domains (as PBR does), competition occurs between nuclear and membrane localizations of the fusion proteins; this may dramatically decrease the sensitivity of the two-hybrid assay. The use of the entire PBR sequence as a bait for the two-hybrid screening was impossible due to the particularly high hydrophobic degree of the protein. To circumvent this problem, we designed a truncated bait protein with only the C-terminal end of the receptor. We recently established that the C-terminal sequence of PBR was the only moiety of the protein to be exposed at the outer side of the mitochondrial membrane and accessible for an eventual partner (25,26). The two-hybrid screening resulted in the identification of a novel protein that specifically interacts with PBR. We named this protein PRAX-1, for peripheral benzodiazepine re-ceptor-associated protein 1. Here, we report first the identification and then the characterization of PRAX-1.

EXPERIMENTAL PROCEDURES
Plasmids and DNA Constructs-The C-terminal end of PBR (aa 156 -169) was used as a bait in the two-hybrid screening. The coding sequence for the last 14 amino acids of PBR was subcloned into the yeast two-hybrid expression vector pGBT9 encoding the GAL4 DNAbinding domain. A double and a triple repeat of the C-terminal motif of PBR spaced with 2 glycine residues was also inserted in pGBT9 to design (CtPBR) 2 and (CtPBR) 3 , respectively. GST-(CtPBR) 3 fusion protein was created by the insertion of the CtPBR 3 coding sequence into the pGEX4T2 vector (Amersham Pharmacia Biotech), in frame with GST. The authenticity of all constructs was verified by sequencing and tested for in-frame protein expression by Western blotting.
Yeast Two-hybrid Screening-All yeast cultures were grown in standard liquid or on solid media, either based on rich YPD medium (1% (w/v) bacto-yeast extract, 2% (w/v) bacto-peptone, 2% (w/v) glucose), or minimal SD medium (0, 67% yeast nitrogen base without amino acids (Difco), 0,2% (w/v) Dropout solution (lacking amino acids involved in the selection desired), 2% (w/v) glucose). For the yeast two-hybrid screening, transformations were performed by the lithium acetate method (27). The Saccharomyces cerevisiae strain HF7c (MATa,112, gal4 -542, gal80 -538, LYS2::GAL1-HIS3, URA3::(GAL4 17mers) 3 -CYC1-lacZ) was first transformed with pGBT9-bait plasmid constructs. Colonies of these transformants were confirmed as His Ϫ and LacZ Ϫ to ensure that the bait alone does not contain transcription activity in HF7c. Different Matchmaker human cDNA libraries purchased from CLONTECH were screened according to the two-hybrid strategy (brain, fetal brain, lung, testis, lymphocytes, leukocytes). Library cDNA from colonies that were His ϩ , LacZ ϩ were isolated and introduced into Escherichia coli MC1061 by electroporation. Purified pACTII plasmids were retransformed alone with pGBT9, with pGBT9-bait, with SNF1 or with lamin gene as control plasmids. The library clones that activate the lacZ reporter gene only in the presence of pGBT9-bait were chosen for sequencing. Both strands sequencing was conducted using the ABI PRISM TM dye terminator cycle sequencing ready reaction kit.
␤-Galactosidase Activity Analysis-␤-Galactosidase reporter activity was determined by plate or liquid culture assays. Plate assays were performed as described elsewhere (28). Positive blue colonies appeared in 30 min to 3 h. In liquid culture assays, ␤-galactosidase activity was determined with chlorophenol red ␤-D-galactopyranoside (Boehringer Mannheim) as substrate according to the procedure of Miller (29). Values of ␤-galactosidase activity reported are the average of triplicate assays of three independent transformants.
In Vitro Protein-Protein Interaction Assay: GST Pull-down-Cultures of E. coli (BL21) were transformed with pGEX4T2 recombinants. Fusion proteins were purified using glutathione-Sepharose 4B according to the manufacturer's instructions (Bulk GST Purification Module, Amersham Pharmacia Biotech). Expressed proteins were identified as GST-(CtPBR) 3 using the 8D7 antibody specific for PBR. For the association studies, binding of GST-(CtPBR) 3 and labeled PRAX-1 was assessed by incubating GST-(CtPBR) 3 adsorbed to glutathione-agarose beads for 2 h at 4°C with [ 35 S]methionine-labeled PRAX-1 protein.
After washing to remove unbound proteins, bound proteins were eluted from the resin by boiling the samples in SDS-polyacrylamide gel electrophoresis gel loading buffer and resolved on a 4 -20% polyacrylamide gel. After electrophoretic separation, gels were dried and exposed to x-ray films.
Cloning of the Entire PRAX-1 cDNA-The human frontal cortex ZAP cDNA library (Stratagene) was screened with the clone resulted from the two-hybrid screening labeled with 32 P using the RTS Rad Prime DNA labeling system (Life Technologies, Inc.). DNA from positive clones was sequenced and then used for the next screening. To obtain the full-length coding sequence, the same library was rescreened using the subsequently isolated clones as probe.
In Situ Hybridization and Fluorescence in Situ Hybridization (FISH) Detection-The chromosomal assignment for the human PRAX-1 gene was performed by in situ hybridization and FISH detection according the procedure of Heng et al. (30,31). The gene mapping was carried out on chromosome preparations obtained from phytohemagglutinin-stimulated human lymphocytes cultured for 72 h. The lymphocyte cultures were synchronized with 5-bromodeoxyuridine (0.18 mg/ml) treatment. The cDNA probe used was of 3.5 kb and comprised the two-hybrid portion.
Determination of the Interaction Domain with PBR in PRAX-1-To determine the interaction domain with PBR in PRAX-1, we used the two-hybrid approach. PRAX-1 deletion mutants were tested for their interaction with PBR qualitatively by a culture test for HIS3 expression and by a colony lift assay for ␤-galactosidase activity. Several PRAX-1(2H) portion domains were subcloned in pACTII to generate differential constructions to be tested. The NcoI PRAX-1(2H) fragment from pACTII-PRAX-1(2H) included the N-terminal 206 aa of the two-hybrid insert. The XmnI/HpaI and BsrBI/HpaI PRAX-1(2H) fragments were also used to test the C-terminal 205 and 118 aa of the PRAX-1(2H) protein, respectively. A truncated version of the PRAX-1(2H) protein was generated by cloning the Cac8I-XbaI PRAX-1(2H) fragment into pACTII. This truncated construction encodes a PRAX-1 fragment from aa 1578 to 1834. A synthetic construction was also designed to fuse two PRAX-1 domains: from aa 1631 to 1656 and from aa 1770 to 1794, linked with three glycine residues.
In Vitro Translation of PRAX-1-PRAX-1 library insert was excised from pACTII by BglII/XhoI digestion and subcloned into pBluescript II KS ϩ into BamHI/XhoI sites. In vitro transcription-translation of the PRAX-1 containing vector was performed in the transcription/translation T-coupled reticulocyte lysate system (Promega) according to the manufacturer's instructions. The [ 35 S]methionine-labeled proteins were fractionated by SDS-polyacrylamide gel electrophoresis before autoradiography or transfer onto nitrocellulose and immunoblotting.
Stable Transfection-Jurkat cells were transfected by electroporation using the Bio-Rad Gene Pulser as described in Ref. 9. Briefly exponentially growing cells (10 7 ) were harvested, washed twice in icecold PBS, and resuspended in 500 l of PBS. The cell suspension was transferred in a 0.4-cm electroporation cuvette and mixed with 40 g of the expression vector containing the human PBR cDNA (pH␤APR-1 neo-hPBR (9)). The cuvette was maintained on ice for 10 min before electroporation (250 miocrofarads, 320 V). After a 10-min incubation at room temperature, the cell suspension was transferred to a 50-ml tissue culture flask. The following day, live cells were counted and plated at 5 ϫ10 4 cells/ml in the enriched medium supplemented with 600 g/ml G418 in 24-well microplates. The selective medium was changed after 48 h. Cells were screened 4 weeks later for the expression of PBR.
Relative Expression Level Analysis-Human RNA Master Blot TM (CLONTECH) was used to analyze relative expression levels of PRAX-1 mRNA in 50 different human tissues and different developmental stages. Hybridization and probe construction were performed as described previously for the MTN blots above.
Immunoblotting-The expression of PRAX-1 protein was analyzed by Western blotting using Human Protein Medley TM (CLONTECH) or human cancer cell line lysates. Cells for protein analysis were lysed in Laemmli buffer, sonicated, and boiled at 100°C for 10 min. Lysates were resolved by SDS-PAGE (4 -12% acrylamide) and electroblotted onto nitrocellulose. Anti-PRAX-1 antibody was produced against aa 1772-1792 of PRAX-1 (Neosystem, Strasbourg, France). Rabbit serum was purified by immunoaffinity on an Affi-Gel column (Bio-Rad) to which the peptides were covalently coupled. Purified antibodies were tested for specific recognition by peptide competition (10 g/ml) before immunoblotting and used at a 1:750 dilution. For subcellular analysis of the expression of PRAX-1, mitochondria were prepared as described in Ref. 32.
Fluorescence Microscopy-The lymphoma cell line U937 was used for immunofluorescence analysis of the PRAX-1 protein subcellular localization. Cells were fixed overnight with 1% paraformaldehyde, washed once and permeabilized for 10 min in a 0.1% saponin, 1% BSA, PBS solution. To visualize PRAX-1, cells were simultaneously incubated with a 1:200 dilution of the rabbit anti-PRAX-1 Ab and mouse antimitochondria M117 monoclonal antibody (Leinco Technologies Inc., St. Louis, MO). Other specific organite markers were also used; specific markers for the nuclear envelope and the endosomes were used according to the manufacturer's instructions (Leinco Technologies, Inc.). After two washes, Cy5-conjugated goat anti-rabbit IgG Abs (Southern Biotechnology Inc., Birmingham, AL) and rhodol green-conjugated goat anti-mouse IgG Abs (Molecular Probes Inc., Eugene, OR) provided fluorescent second step reagents for rabbit and mouse Abs, respectively. Analysis of the subcellular distribution of the PRAX-1 protein distribution was performed with a laser scanning confocal microscope (LSM 410, Zeiss, Oberkochen, Germany) equipped with a c-Apochromat water immersion lens (ϫ63, NAϭ1.2). Specificity controls were carried out by preincubation of anti-PRAX-1 Abs with the immunizing peptide at 10 g/ml.
Immunocytochemical and Immunofluorescence Analysis of the PRAX-1 Protein Expression in the Rat Brain-Immunocytochemical localization of the PRAX-1 protein was carried out in a series of sections prepared from brain rats. Briefly, 50-m brain sections were incubated for 48 h at 4°C with polyclonal anti-PRAX-1 antibody diluted 1:800 with PBS containing 2% BSA and 0.1% Triton X-100. The following day, the sections were washed with PBS, labeled with the secondary antibody, a peroxidase-conjugated swine anti-rabbit IgG (DAKO, diluted 1:500), and the sections processed for immunocytochemistry analysis. Staining was developed in 0.02% hydrogen peroxide in PBS. The reaction was stopped by distilled water. Sections were mounted on gelatincovered slides, air-dried, and analyzed using a Leica DMLB microscope. For immunofluorescence analysis, the sections were labeled with the secondary antibody, a CY3-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, diluted 1:200). Primary and secondary antibodies were diluted in PBS containing 1% normal goat serum, 2% bovine serum albumin, and 0.1% Triton X-100. Immunostained sections were mounted in Mowiol (Calbiochem, La Jolla, CA) and observed with a laser scanning confocal microscope (LSM 410, Zeiss).
Sequence Analysis-An Applied Biosystems model 373A sequencer and the ABI PRISM TM dye terminator cycle sequencing ready reaction kit were used for sequencing. Sequence analyses were performed using the University of Wisconsin Genetics Computer Group package (GCG; Ref. 33). The percentage of similarity and pairwise homologies were assessed using the Bestfit software. Data base searches were conducted against GenBank TM and European Molecular Biology Laboratory releases.

Isolation of a Protein That Interacts with PBR
A two-hybrid screening was conducted to identify proteins that interact with PBR. The PBR protein resides in the outer mitochondrial membrane with the bulk of the protein integrated into the membrane and the C-terminal domain oriented toward the cytosol. The last 14 amino acids of the protein were thus the longest hydrophilic portion of the protein demonstrated as accessible with the monoclonal 8D7 anti-PBR antibody specifically recognizing this portion (25). Therefore, the first bait we used was the C-terminal end of PBR, the last 14 amino acids referred as CtPBR (aa 156 -169). Libraries of hybrid proteins between the GAL4 activating domain and random cDNA fragments derived from human brain, fetal brain, lung, testis, lymphocytes, or leukocytes were first screened. From about 30 ϫ 10 6 transformants sequentially screened, no specific clones were identified. Although the expression and the nuclear localization of the bait proteins were confirmed by Western blot analysis and confocal microscopy using the anti-PBR antibody 8D7 (data not shown), we failed to detect any specific clones interacting with the bait. One explanation would be the very short length of the bait we used. We therefore constructed a longer bait with three repeats of the C-terminal motif of PBR spaced with two glycine residues, (CtPBR) 3 . Then, the HF7c strain containing pGBT9-(CtPBR) 3 was transformed with a library of human fetal brain cDNA fragments expressed as fusions to the GAL4 activation domain. Screening of about 2 ϫ 10 6 transformants resulted in 411 His ϩ clones; 10 of these clones were ␤-galactosidase-positive. These 10 plasmids that activated both HIS3 and LacZ genes were tested for specificity by pairing them with a plasmid encoding a hybrid of the Gal4 DNA-binding domain with a protein unrelated to PBR: SNF1 or lamin gene. Four of these 10 clones encode proteins that specifically interact with our (CtPBR) 3 bait only (Fig. 1). Sequence analysis revealed that these four clones were identical; the library clones isolated contained the same 2152-bp insert having a poly(A) tail. As assessed by ␤-galactosidase activity test and measurement, the four clones isolated strongly interact with the triple repeat of the C-terminal end of PBR; weaker interaction was detected when a double repeat of the C-terminal motif replaced the triple one; and no interaction occurred between the four clones and CtPBR (Fig. 1).

The cDNA Isolated Encodes a New Protein
The 2152-bp insert isolated from the two-hybrid screening contained a poly(A) tail and a polyadenylation signal AATAAA occurring 12 nucleotides upstream from the poly(A) tail. The only translation of the insert in-frame with the GAL4 activating domain was a 294-amino acid protein. The stop codon was located 1223 nucleotides upstream from the poly(A) tail. The 3Ј-untranslated region included an ATTTA site, which has been associated with mRNA destabilization in cytokine and growth factor transcripts (34). Data bank comparison searches indicated that it encodes a novel protein never described before. We named it PRAX-1, for peripheral benzodiazepine receptorassociated protein 1. As the insert lacks the methionine initiation-codon ATG at its 5Ј end, we then isolated a portion of the PRAX-1 protein (Fig. 2). This PRAX-1 portion, referred as PRAX-1(2H), is a polypeptide of 32 kDa. The full-length cDNA of the protein was obtained from the sequential screening of a human brain ZAPII cDNA library with different 5Ј PRAX-1 cDNA probes. The complete PRAX-1 cDNA is 7033 bp and contains an open reading frame of 5571 bp. The first in-frame ATG codon, located at nucleotide 198, matched the Kozak consensus motif (35) and was followed by a stop codon at position 5768. Supporting the assignment of the initiating amino acid as the first methionine, there is an in-frame termination codon (TGA) located 42 base pairs from the initiating codon. This open reading frame encodes a predicted protein of 1857 amino acid residues with a calculated molecular mass of 200 kDa and an acid pI of 4.93. The two-hybrid insert extends from base 4887 to 7033, and the corresponding PBR-interacting portion of PRAX-1, PRAX-1(2H), extends from aa 1564 to 1857 (Fig. 2). The amino acid sequence analysis showed that PRAX-1 exhibits particularly high percentages of proline, glutamic acid and leucine residues: 10.6%, 10.7%, and 10%, respectively. A Kyte-Doolittle hydropathy plot of the deduced amino acid sequence reveals a rather hydrophilic protein with few hydrophobic domains but no typical transmembrane domain or hydrophobic leader sequence. Sequence analysis using the Prosite software did not identify any functional signature but revealed the presence of several protein motifs such as 2 potential N-glycosylation sites, 2 amidation sites, 3 glycosaminoglycan attachment sites, and a high number of putative phosphorylation sites (2 cAMP-and cGMP-dependent protein kinase, 28 protein kinase C, 38 casein kinase II, and 2 tyrosine kinase consensus phosphorylation sites). In addition, we identified by homology searches three sequences containing the 4-aa motif Lys-Arg/ Lys-X-Arg/Lys and a 6-aa motif Arg-X-X (hydrophobe)-X-X-Ser, which represent, respectively, the minimal nuclear localization signal consensus sequence and a mitochondrial targeting signal. PRAX-1 contains two long glutamic acid stretches with successive 13 and 12 residues (aa 1262-1274 and 1334 -1346). The most interesting feature of the sequence was the presence of several different motifs implicated in protein-protein interaction. Indeed, homology searches revealed the presence of three proline-rich domains where the proline content exceeds 20% (residues 528 -597, 23%; residues 1071-1150, 30%; residues 1715-1782, 21%), three leucine zipper motifs at the Nterminal side of the protein (positions 126, 162, and 188), and a SH3-like domain in the PRAX-1(2H) portion from aa 1636 to 1692 (Figs. 3 and 4). This latter exhibits 50 -60% homology with the SH3 domain of the Src protein kinase family and of the adaptor proteins Grb2, Sem-5, and DRK (Fig. 4). Finally, further inspection revealed that the PRAX-1(2H) portion contains FIG. 1. PRAX-1 specifically interacts with PBR in the twohybrid system. A, HIS3 expression. Yeast were co-transformed with the indicated pGBT9 and pACTII constructs and plated on SC-Trp-Leu. After 3 days, a colony was picked that was patched on SC-Trp-Leu-His plates to test for transcriptional activation of the HIS3 gene. B, LacZ expression. Co-transformed yeast as indicated were plated on SC-Trp-Leu-His. After 3 days, a qualitative and quantitative ␤-galactosidase assays were performed as described under "Experimental Procedures." These plasmids contain human Ras and Raf cDNA coding sequences, Ras and Raf interact strongly and serve as a positive control and reference (taken as 100%) for the ␤-galactosidase activity evaluation. ND, not determined. two 25-residue domains exhibiting 68% homology and 64% identity (VALFDYDPVSMSPNPDAGEEELPFRE, from aa 1631 to 1656 and VAAFDYNPQESSPNMDVEAELPFRA, from aa 1770 to 1794). These two domains, which are spaced by about 100 residues, flank the SH3-like domain and the prolinerich portion of PRAX-1(2H) (Fig. 3). Noticeably, these two domains are the most hydrophilic and acidic portion of the PRAX-1(2H), with pI 3.5 and 3.7, respectively. To confirm the interaction observed in yeast, we tested whether in vitro translated [ 35 S]methionine-labeled protein PRAX-1(2H) was able to interact with C-terminal PBR sequences fused to glutathione S-transferase. The expression of the bait GST fusion proteins (GST-(CtPBR) 3

Search for the PRAX-1 Motif Interacting with PBR
To identify the PRAX-1(2H) domain implicated in the interaction with PBR, we tested different PRAX-1(2H) portion constructions for their ability to bind with PBR in the two-hybrid system (Fig. 6). The N-terminal 206 aa of PRAX-1(2H) contains the SH3-like domain and a portion of the proline-rich domain. The C-terminal 205-aa portion of PRAX-1(2H) begins in the middle of the SH3-like domain and contains the entire prolinerich domain. The C-terminal 118-aa portion of PRAX-1(2H) only contains a portion of the proline-rich domain and the C-terminal end of PRAX-1(2H). As shown in Fig. 5, neither the N-terminal 206 aa, nor the C-terminal 205 aa, nor the Cterminal 118 aa of the PRAX-1(2H) gives positive results in the two-hybrid test. None of these constructions is sufficient to interact with PBR. These data suggested that both the SH3like motif and proline-rich domain were required for specific interaction with PBR. In line with these results, the presence of the two repeated regions of 25 residues flanking the SH3-like and the proline rich domains is probably needed for the interaction. To examine the role of this duplicated motif in the interaction with our bait PBR, we then tested a truncated portion of the PRAX-1(2H) lacking its first 13 and its last 23 amino acids and a synthetic construction where the two repeated regions are joined. In a qualitative LacZ expression test, the first construction gives positive results, the latter synthetic region gives weak but positive signal in the two-hybrid test (Fig. 6). As the latter construction was designed with three glycine residues as a linker between the duplicated motif, the weaker signal we observed in the two-hybrid test may be explained by the non-optimal length of the linker. It would then be interesting to test different linker length to optimize the signal. Altogether, these results showed a complex interaction between PRAX-1 and the bait PBR insofar as different PRAX-1 elements clearly took part in the interaction with the bait: the SH3-like domain, the proline-rich domain, and the duplicated motif.

The Human PRAX-1 Gene Is Located on Chromosome 17
The PRAX-1 gene was mapped using FISH on normal human chromosomes. A 3.5-kb portion of the PRAX-1 cDNA was used as a probe. FISH signals and the DAPI banding pattern were recorded separately by taking photographs; the assignment of the FISH mapping data with chromosomal bands was achieved by superimposing FISH signals with DAPI-banded chromosomes. In 100 metaphase cells examined after in situ hybridization, 73 mitotic figures showed signals on one pair of the chromosomes. DAPI banding identified the chromosome 17 and assigned the probe signal to a single locus, the q22-q23 region of the long arm of chromosome 17. These results mapped the PRAX-1 probe to the 17q22-q23 region of the long arm of the human chromosome 17 (Fig. 7).

PRAX-1 mRNA Tissular Expression
The tissue distribution of PRAX-1 mRNA was first studied by Northern blots. The Northern blots shown in Fig. 8 were probed in parallel with the same preparation of 32 P-labeled PRAX-1 probe and exposed for 24 h. PRAX-1 mRNA was detected as displaying various transcript sizes of 6, 7.5 and 9.5 kb heterogeneously distributed, but the main signal was 7.5 kb found in the human brain (Fig. 8A). To further characterize the distribution of PRAX-1 expression, we examined the relative PRAX-1 mRNA expression in 50 different human tissues and developmental stages (Fig. 8B). In agreement with the previous blot, the brain sample contained the most PRAX-1 mRNA followed by the pituitary gland and thymus. Kidney and ovary displayed significant PRAX-1 mRNA expression. Among human fetal tissues, the highest PRAX-1 mRNA expression level was found in the fetal brain followed by the fetal heart, fetal kidney, and thymus. Focusing on different human brain regions, the temporal lobe and the putamen displayed the highest PRAX-1 mRNA expression level, followed by amygdala, caudate nucleus, cerebral cortex, occipital lobe, and frontal lobe. Low expression levels were found in cerebellum, hippocampus, substantia nigra, thalamus, and subthalamic nucleus. No expression was detected in medulla oblongata and spinal cord.
PRAX-1 mRNA expression was also studied among different species. Polymerase chain reaction studies using different specific primers for the human PRAX-1(2H) were performed with cDNA derived from murine or rat brain, and revealed significant homology among rat, murine, and human PRAX-1. Amplicons were sequenced and found similar to a 80% extent; rat and mouse amplicons were 83% and 81%, respectively, similar to the human amplicon, respectively. At the amino acid level, they exhibit a 90% homology. Such a high homology made it possible to use the human PRAX-1 probe for analyzing the homologous PRAX-1 mRNA expression in rat and mouse. Northern blot hybridization experiments on rat and murine tissues revealed a 7.5-kb mRNA in the murine and rat brain markedly detected after a 6-h exposure. Both the distribution and size of the mRNA transcript observed in mouse and rat are identical to that observed in human (Fig. 8C).

A Protein of the Predicted Molecular Mass Is Expressed in Vivo
In Vitro Test for Specific Anti-PRAX-1 Antibody Recognition-The predicted molecular mass of the two hybrid PRAX-1 portion was first verified by transcription-translation of a pBluescript plasmid containing the library insert in the transcription/translation coupled reticulocyte lysate system (Promega). Proteins synthesized in the presence of [ 35 S]methionine were fractionated in a 4 -20% (w/v) SDS-polyacrylamide gel, and the labeled proteins were visualized by autoradiography. Identification of the produced proteins was also confirmed with specific polyclonal anti-PRAX-1 antibody raised against aa 1774 -1794 (Fig. 9). As expected, the protein produced from the library insert has an apparent molecular mass of 35 kDa in accordance with the calculated mass of 34,680 daltons for the construct.
Expression Pattern of the PRAX-1 Protein-The expression analysis of the PRAX-1 protein was performed using lysates from either normal or tumoral human tissues and cell lines. As shown in Fig. 10, specific anti-PRAX-1 antibody labeling found a single band appearing in the 220 -250-kDa range. Despite a predicted molecular mass of 200 kDa, the PRAX-1 protein migrates as a 220 -250-kDa protein. The difference between the calculated and the SDS-PAGE-determined molecular weight is likely due to the presence of the two highly charged glutamic acid stretches and the acidic nature of the protein (pI 4.93) that may slow down the protein migration. Among the normal human tissues studied, the PRAX-1 protein was found to be predominantly expressed in brain and thymus in accordance with the Northern blots results. According the high homology among human, rat, and mouse PRAX-1 sequences, we have used the anti-PRAX-1 antibody raised against the human protein to analyze the expression profile of the homologous PRAX-1 in rat and murine tissues or cell lines. As expected, we found a single protein migrating in the 220 -250-kDa range both in rat and mouse. Strikingly, all the human cell lines tested expressed the PRAX-1 protein. Interestingly, we noticed a particularly high expression of PRAX-1 in cancer cell lines. In comparison to their normal counterparts, PRAX-1 was indeed found highly expressed in U373 or SH SY 5Y cells versus brain, Jurkat, or Molt-4 cells versus thymus or SW480 cells versus colon, HA-CAT, SVK14, or NCTC 2544 versus skin. All the cancer cell lines tested expressed PRAX-1 protein in larger amounts than their respective counterparts. The PRAX-1 protein was also found expressed in cancer cell lines for which the normal counterparts were not demonstrated to express PRAX-1. These results clearly demonstrated the expression of PRAX-1 as a single protein migrating at 220 -250 kDa and suggested a differential expression of the PRAX-1 protein associated with cell proliferation.

PRAX-1 Subcellular Localization
The cellular distribution of the PRAX-1 protein was examined using Western blotting, immunochemistry, and immunofluorescence. Western blotting performed on isolated mitochondria revealed the 220 -250-kDa band for PRAX-1. Specific labeling was ascertained with competition of the corresponding immunogen peptide (Fig. 11). Different mitochondria preparations were tested, and interestingly different PRAX-1 expression levels were observed with the preparation origin. Although U937 and U373 mitochondria clearly exhibit the 220 -250-kDa PRAX-1 protein, the PRAX-1 protein was obviously found less abundant in Jurkat mitochondria. As the Jurkat was the only human cell line described devoid of PBR (9), we wondered whether the PRAX-1 mitochondrial subcellular localization was driven by PBR. We therefore tested the expression of the PRAX-1 protein in mitochondria prepared from transfected Jurkat cells expressing PBR (Fig. 11). Strikingly, the expression level of the PRAX-1 protein was found to be much higher in mitochondria from PBR-transfected Jurkat mitochondria than that in wild type Jurkat mitochondria. These results indicate that the mitochondrial subcellular localization of PRAX-1 was dependent on the presence of PBR.
Confocal microscopy analysis was performed on the lym-phoma cell line, U937. As shown in Fig. 12, the protein appeared cytoplasmic, being mostly absent in the nucleus. By immunofluorescence analysis, a punctate vesicular pattern was observed. Colabeling experiments with a specific marker for mitochondria indicate that the PRAX-1 protein is cytoplasmic and partially colocalizes with the mitochondria (Fig. 12A). Specific markers for other subcellular structures were used and did not show any other colocalization; as shown in Fig. 12 (B  and C), the PRAX-1 labeling did not colocalize with the nuclear envelope or with endosome markers.
FIG . 8. Expression of PRAX-1 mRNA. A, expression of PRAX-1 mRNA in various human tissues. Northern blots containing 2 g of poly(A) ϩ RNA isolated from the indicated tissues were probed with a 837-bp PRAX-1 probe as described under "Experimental Procedures." The autoradiogram was exposed for 24 h. B, relative expression of PRAX-1 mRNA in 50 different human tissues and developmental stages. The autoradiogram was exposed for 18 h. The diagram on the rigIht shows the nature and the position of poly(A) ϩ RNAs and controls. C, expression of PRAX-1 mRNA in various rat and murine tissues. Northern blots containing 2 g of poly(A) ϩ RNA isolated from each indicated tissues were probed with a 837-bp PRAX-1 probe as described under "Experimental Procedures." The autoradiograms were exposed for 24 h.

Immunocytochemical Analysis of PRAX-1 Protein Expression in Rat Tissues
Various rat tissues were examined immunocytochemically with anti-PRAX-1 antiserum as a probe. Our polyclonal anti-PRAX-1 antibody raised against the human protein cross-re-acting with the rat PRAX-1 homologue made it possible to study the expression of the PRAX-1 protein in different rat brain regions and peripheral tissues.
Among the tissues examined with the PRAX-1 antibody, the brain sections exhibited the most intense staining. Focusing on the brain, we found selective and restricted labeling with intense signals observed in the nucleus accumbens (shell), lateral septal nucleus, paraventricular and supra-optic hypothalamic nuclei, CA1 subfield of the hippocampus, amygdala, piriform cortex, cingulate cortex, and the olfactory bulb. As shown in Figs. 13 and 14, we observed in coronal and sagittal sections of the rat brain a marked labeling of PRAX-1 restricted to some brain areas. In the hippocampus, the labeling was restricted to the pyramidal cell layer of the CA1, being absent in the CA2 and CA3. The labeling is neuronal, intense in the cell cytoplasm, the proximal axonal processes also exhibit high PRAX-1 immunoreactivity (Fig. 15).

DISCUSSION
In our attempt to further characterize the peripheral benzodiazepine receptor general function, we decided to use the yeast two-hybrid system to identify protein(s) that specifically interact(s) with this receptor. Since PBR exhibits a particularly high hydrophobic degree with five transmembrane-spanning domains, we used the C-terminal end of the protein as the bait in our screening. Recently, we produced a monoclonal antibody against the C-terminal 12 amino acids of PBR that corresponds to the hydrophilic portion of the protein located outside the mitochondrial membrane (25). Our choice of the C-terminal FIG. 9. In vitro expression of the PRAX-1 protein. A, transcription-translation of the pBluescript PRAX-1-containing vector was performed in the transcription/translation-coupled reticulocyte lysate system (Promega) in the presence of [ 35 S]methionine. The labeled proteins were fractionated by SDS-polyacrylamide gel electrophoresis before autoradiography. PRAX-1(2H), translational products of the two-hybrid screening insert, and control, translation products obtained from a pBluescript vector without insert. B, Western blot of translational products of PRAX-1 cDNA. Translated proteins were labeled with specific anti-PRAX-1 antibody as described under "Experimental Procedures." Competition was performed with 10 g/ml immunogen peptide.
FIG. 11. Western blotting analysis of PRAX-1 expression on mitochondria. Mitochondria were prepared as described under "Experimental Procedures," electrophoresed on a 4 -12% acrylamide gel, and transferred on nitrocellulose membrane before immunolabeling with specific PRAX-1 antibody. Competition with 10 g/ml immunogen peptide was performed in parallel as a control to ascertain the specificity of the labeling observed. end of PBR as a bait in our two-hybrid screen was driven by this specific anti-PBR antibody, clearly demonstrating that this portion of the PBR protein is accessible for an eventual partner of PBR. We first performed a two-hybrid screening with a bait protein fusion designed with aa 156 -169 of PBR fused to the GAL4-binding domain. Libraries we screened were chosen according to the well known tissular PBR expression. Lung, testis, brain, fetal brain, lymphocyte, and leukocyte cDNA libraries were screened for partners that interact with the 14-amino acid C-terminal motif of PBR, without success; no specific clones were isolated. We then decided to design a two-hybrid bait that consists of repeats of the C-terminal motif of PBR. We chose to use a triple repeat as a bait because it enabled a correct folding of the peptide and allowed placement of the motif at varying distances from the GAL4-binding domain. We thought that the use of a triple repeat improved the chances of identifying putative interacting proteins. In practice, we isolated clones that give strong positive signals with the triple motif of the bait and weaker signals with the double motif. The clones we have isolated do not interact with the monomere. This strategy based upon peptide-protein interactions made it possible for us to trap a new protein that specifically interacts with PBR. Such an approach would prove particularly efficient for screening interacting partners with the two-hybrid procedure, even if the bait protein exhibits high hydrophobic degree and few or small hydrophilic portions. As an example, Ohno et al. (36) used this strategy of a multimeric bait to identify clathrin-associated proteins (AP-1, AP-2) as interacting with the tyrosine-based sorting signals of several integral membrane proteins.
We have cloned a cDNA coding for a human protein that interacts with the peripheral benzodiazepine receptor. The cDNA we isolated has no homology to any known proteins in current data banks. We named this new protein PRAX-1, for peripheral benzodiazepine receptor-associated protein 1. Sequence analysis of the PRAX-1 cDNA revealed the presence of an open reading frame of 5571 bp encoding a 200-kDa polypeptide of 1857 amino acids. The portion of PRAX-1 trapped in the two-hybrid screening is the C-terminal 294 aa. To elucidate the functions of this new protein, we searched the Prosite data base for specific protein function signature; none was found. The most characteristic feature of the PRAX-1 protein was the presence of several different motifs implicated in protein-protein interactions; sequence analysis and homology searches indeed revealed 3 proline-rich domains, 3 leucine zippers, and a SH3-like domain. The latter is a 60-amino acid region that exerts 60% homology with the Src family protein kinase SH3 domain (37) and with the SH3 domain of adaptor proteins mediating binding of guanine nucleotide exchange factors to growth factor receptor, GRB2 (38), , and DRK (40). The function of the SH3 domains is not well understood, but they are known to bind to proline-rich ligands. Interestingly, focusing on the two-hybrid insert, one of the three stretches of proline residues between aa 1718 and 1782, where the proline rate reaches 21%, is located downstream of the SH3-like domain.
To supplement the yeast two-hybrid results, we have tested the PRAX-1/PBR interaction using the GST pull down technique. The high hydrophobic degree of the PBR protein, prevented us from using the whole protein for this interaction test. We then tested the interaction of PRAX-1 with the bait PBR used in the two-hybrid trap. Our results clearly indicated that the PRAX-1 portion we isolated in the two-hybrid screen interacted in vitro with our bait PBR. Immunoprecipitation studies were also impaired due to the high hydrophobic degree of the PBR protein. Indeed, the conditions we were compelled to use to solubilize the membraneous PBR failed to show the interaction between PRAX-1 and PBR in intact cells. Indeed, all the buffers we tested (deoxycholic acid, Nonidet P40 or Triton X-100-based buffers) were stringent enough to solubilize PBR, but they were too stringent to allow antibody binding or protein-protein interaction.
PRAX-1 mRNA shows three transcript sizes of 6, 7.5, and 9.5 kb and has a tissue-specific distribution. The predominant species was found to be the 7.5-kb species, mainly expressed in the brain. Several brain regions distinguished with particularly high mRNA content: temporal lobe, frontal lobe, cerebral cortex, putamen, and hippocampus. Using Western blotting and specific anti-PRAX-1 antibody, PRAX-1 was detected as a single band as a 220 -250-kDa protein. The expression profile of the protein confirmed the expression profile defined for the mRNA messenger. Thus, PRAX-1 was found to be expressed in the human brain and thymus, as well as in the studied human cell lines (either normal or tumoral). PRAX-1 expression analyzed in several mouse and rat samples matched that of the human PRAX-1 both at the mRNA and the protein levels. Very interestingly, the expression pattern reveals higher expression levels of the PRAX-1 protein in cancer cell lines versus normal tissues. We indeed showed that tumoral tissues exhibit higher expression levels; although some cancer cell lines expressed high amount of PRAX-1 protein, their normal counterparts do not exhibit detectable PRAX-1 expression. Our observations showed a differential expression for the PRAX-1 protein associated with cell proliferation.
A comparative analysis of the PRAX-1 and PBR expression, both at the mRNA and protein levels (data not shown), shows that PBR exhibits a rather broader expression profile than PRAX-1 does. Indeed, some tissues do express PBR but seem to be devoid of any detectable PRAX-1 expression. This may indicate that PRAX-1/PBR interaction is not an exclusive partnership and suggests that other PRAX-x exist.
PRAX-1 subcellular distribution analyses not only reveal a mitochondrial localization for PRAX-1, they also show a cytoplasmic presence of the protein. We found mitochondrial localization for the PRAX-1 protein on Jurkat, U937, and U373 cell lines. Very interestingly, the expression pattern differed according to the cell line tested; indeed, mitochondria from the Jurkat cell line were found to exhibit a very low PRAX-1 expression level compared with the other mitochondrial preparations tested. To date, the Jurkat cell line is the only cell line described devoid of any PBR expression (9); the prevailing question was then to test the role of PBR in the PRAX-1 localizing on mitochondria. We therefore analyzed the PRAX-1 expression in Jurkat cell line transfected with PBR. Remarkably, we found that the expression level of the PRAX-1 protein in mitochondria from transfected Jurkat cells expressing PBR was much higher than that observed in Jurkat WT. In the presence of PBR, we showed that the PRAX-1 protein was found preferentially expressed on mitochondria. Our results, clearly indicating that the PRAX-1 subcellular localization was dependent on the presence of PBR, show that the PRAX-1 localizing on mitochondria is PBR-driven.
Immunocytochemistry was carried out to define the distribution of the PRAX-1 protein in the rat brain. Our results clearly showed a neuronal expression of the PRAX-1 protein. This may be in contrast with the PBR protein being mostly described to be glial. To date the expression of the PBR in the brain has been mostly performed using radiolabeled specific PBR ligands. These experiments have clearly demonstrated that the PBR is expressed on glia. However, we think that such a technique may suffer from a lack of sensitivity because of the low basal expression level of PBR in the brain. Most studies that showed PBR expression in the brain have been performed following brain damages that induced gliosis, leading to the up-regulation of the PBR expression. The dogma of the restricted expression of the PBR in glia may be discussed considering a previous observation by Anholt et al. (41), who showed that PBR is also expressed in nerve terminals in the olfactory bulb. To date, no complete expression distribution analysis using specific anti-PBR antibody has been undertaken. Such an analysis is warranted to determine whether PBR is exclusively glial or glial and neuronal. The latter hypothesis and PRAX-1 being a rather brain-specific protein would raise the question of the PRAX-1 protein implicating PBR in a specific function that must be unraveled.
In the brain, the restricted pattern of the staining we observed was most prevalent in the mesolimbic system. The rat PRAX-1 protein was indeed obviously expressed in the CA1 subfield of the hippocampus, in the nucleus accumbens, lateral septal nucleus, paraventricular and supra-optic hypothalamic nuclei, amygdala, piriform cortex, cingulate cortex and the olfactory bulb. The mesolimbic system has been described critical for goal-directed behaviors and pathophysiologically associated with neuropsychiatric disorders including dementia, FIG. 13. Immunocytochemical staining of rat brain sections with anti-PRAX-1 antibody: sagittal sections. Sections of rat brain were immunostained with the polyclonal anti-PRAX-1 antibody as described under "Experimental Procedures." In A, a whole sagittal section is shown; in B, a sagittal section focused on the CA1 subfield of the hippocampus (ϫ 20) is shown, which illustrates the restricted labeling of the CA1, as opposed to CA2 and CA3. schizophrenia, and affective disorders. To date, the aspects of behavior mediated or the mechanism of their modulation by this system remain a matter of conjecture. Analyzing the expression of the PRAX-1 protein and its modulation in this system may bring further information. Using a polyclonal anti-PRAX-1 antibody, we have thus shown a specially restricted expression pattern for the PRAX-1 protein in the brain. Studies are carried out to determine whether the expression profile we observed either resulted from a real restricted expression of the protein or illustrated a special labeling that occurred following change in the antigenic epitope: this epitope targeted by the polyclonal antibody may be masked in some tissues and accessible in others (following phosphorylation for example).
Furthermore, using a 3.5-kb portion of the PRAX-1 cDNA as a probe in FISH analysis, the PRAX-1 gene was unambiguously located in the q22-q23 region of the human chromosome 17. Among other genes located near the PRAX-1 region on the long arm of chromosome 17 are the homeobox B cluster, the coilin p80, the tumor necrosis factor-␣-induced protein 1, the protein kinase C polypeptide, or the growth hormone/placental lactogen gene cluster. Interestingly, different inherited or acquired diseases have been described implicating the long arm of the chromosome 17, in particular hereditary degenerative dementia (frontotemporal lobe dementia as an example; FLDEM, Ref. 42). Deciphering the DNA defects involved in those diseases may then be facilitated by the availability of this new cloned probe for the distal region of this chromosome.
The structural features of the PRAX-1 protein suggest a modular model for its interaction with PBR. As previously mentioned, the PRAX-1 protein portion isolated in the twohybrid screen contains a proline-rich domain and a SH3-like  domain. Our studies demonstrated that the interaction of PRAX-1 with PBR requires both the proline-rich domain and the SH3-like domain. As the proline rich stretch contains a PPKPRR motif that belongs to the class II ligands for the SH3 domains (PXXPXR) we assume that this region may be recognized as a target by the SH3-like domain (43,44). These interacting regions may be responsible for a folding of PRAX-1 that brings closer two other regions of PRAX-1. Indeed two 25amino acid regions showing significant similarity with 68% homology and 64% identity flank the SH3 domain and the proline-rich stretch. These repeated sequences coincided with the most hydrophilic and acidic regions of the two-hybrid PRAX-1 insert, and thus are candidates to be the contributing PRAX-1 regions for the interaction with PBR. The PBR baitbased motif has a basic pI of 11.8. Ionic forces and electronic interactions would stabilize the interaction. To challenge this interacting scheme, we tested different portions of PRAX-1(2H) for interacting with the bait PBR in the two-hybrid system. All the constructions that contain only portions of either the SH3like domain or the proline-rich domain failed in interacting with the bait PBR, evidencing their common important role in contributing to the interaction with PBR. PRAX-1 was trapped according to the two-hybrid strategy as interacting with a repeated motif based on PBR. We thus assume that the PRAX-1/PBR interaction is modular with a single PRAX-1 protein interacting with several PBR molecules. A structural model has shown that PBR was organized in clusters of 4 -6 molecules (45). Such a multimeric topography makes it possible for the PRAX-1 protein to cover the PBR molecules via its interaction with the C-terminal end of the PBR protein (Fig. 16).
As the PRAX-1 protein exhibits several motifs implicated in protein-protein interaction, PRAX-1 may act as an adaptor protein to recruit different targets to get them in the vicinity of PBR. PRAX-1 linking PBR to other cytoplasmic effectors must be considered. PRAX-1 is the first cytoplasmic protein described as interacting with the mitochondrial PBR. PRAX-1 is a new protein that has never been described before. Thus, PRAX-1 opens a new area of investigation for PBR and may be considered as the link that was lacking for better understanding of PBR. Although PBR was restricted to the mitochondria, PRAX-1 exhibits larger localization and may link PBR to either cytoplasmic or nuclear effector that must be identified. FIG. 16. Hypothetical model of the interaction between PBR and PRAX-1. Schematic diagram of the proposed intermolecular interactions between PBR and PRAX-1. PRAX-1 as a monomer associates with several molecules of PBR (at least two of them) recruiting them in its vicinity.